© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 1-RES-SM
    11. 28260
    Prev
    Next
    INDEX
    Explanations

    references to champions and competitions in combat sports

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    ctigges/pythia-70m-deduped__res-sm_processed/1-res-sm
    Prompts (Dashboard)
    32,768 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.1.hook_resid_post
    Hook Layer
    1
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    341
    +0.16
    0.9%
    481
    +0.13
    0.7%
    169
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    111
    +0.16
    0.10
    156
    +0.13
    0.07
    271
    +0.12
    0.02
    Negative Logits
     Redistributions
    -1.56
    ↵Č
    -1.54
     large
    -1.52
    ke
    -1.47
    uto
    -1.47
    ices
    -1.46
    tering
    -1.43
    ago
    -1.41
     canal
    -1.39
     reading
    -1.38
    POSITIVE LOGITS
    fight
    1.93
     champion
    1.87
     champions
    1.81
     MMA
    1.80
     Champions
    1.78
     UFC
    1.76
     Fight
    1.71
     contender
    1.70
     supremacy
    1.69
     finale
    1.67
    Activations Density 0.809%

    No Known Activations